Wolfram LLM Benchmarking Project: leaderboard
Models turn English-language specifications into Wolfram Language code, using exercises from Stephen Wolfram's An Elementary Introduction to the Wolfram Language as the test cases. Wolfram runs the generated code through its own tools for checking functional correctness and publishes two columns, valid syntax and correct functionality. This board reports the correct functionality percentage, higher is better.
Metric: Correct Functionality (%). Source: www.wolfram.com. Saturation forecast: Around January 2027. 598 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Fable 5.1 (High) | 75.1 | #9 (Claude Fable 5.1) |
| 2 | Gemini 3.8 Flash (Medium) | 73.8 | #21 (Gemini 3.8 Flash) |
| 3 | Gemini 3.7 Flash (Medium) | 73.6 | #23 (Gemini 3.7 Flash) |
| 4 | Gemini 3.8 Flash (High) | 73.3 | #21 (Gemini 3.8 Flash) |
| 5 | Gemini 3.7 Flash (High) | 73.3 | #23 (Gemini 3.7 Flash) |
| 6 | Grok 4.6 (High) | 72.9 | #26 (Grok 4.6) |
| 7 | Claude Opus 4.7 (Thinking) | 72.5 | #49 (Claude Opus 4.7) |
| 8 | Gemini 3 Pro (Preview) | 71.2 | #66 |
| 9 | Gemini 3.1 Pro (Preview) | 71.1 | #58 |
| 10 | Grok 4.6 (Low) | 71.1 | #26 (Grok 4.6) |
| 11 | Gemini 3.6 Flash (Medium) | 70.9 | #51 (Gemini 3.6 Flash) |
| 12 | Gemini 3.6 Flash (High) | 70.8 | #51 (Gemini 3.6 Flash) |
| 13 | Gemini 3.5 Flash (Medium) | 70.3 | #60 (Gemini 3.5 Flash) |
| 14 | Grok 4.5 (High) | 70 | #48 (Grok 4.5) |
| 15 | Grok 4.7 (High) | 69.9 | #25 (Grok 4.7) |
No result here: #8 Gemini 4 Argon, #10 Muse Wonder Full.
Interactive version: theaggregate.ai/benchmark?slug=wolfram-llm-benchmarking-project · How It Works · Data refreshed daily, snapshot 2026-10-20.